Qualitative reasoning about control

Qualitative models are better suited for some tasks than the traditional quantitative, or numerical models. These tasks include diagnosis, generating explanation of the system's behavior and designing novel devices from first principles. It is shown by an example how a qualitative modeling approach can be applied to the control of a dynamic system. The example used is the balancing of a pole on a cart. A control rule is derived from a qualitative model of the pole-and-cart system. This rule is shown to be a straightforward qualitative abstraction of the "classical" control rule derived from the differential equations model of this system.<
>